{
  "best_global_step": 1126,
  "best_metric": 1.2738397121429443,
  "best_model_checkpoint": "finetuned_models/FFT/checkpoint-1126",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1689,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 1.5691047963500022,
      "epoch": 0.044444444444444446,
      "grad_norm": 61.75,
      "learning_rate": 2.840236686390533e-06,
      "loss": 2.82507080078125,
      "mean_token_accuracy": 0.5578909605741501,
      "num_tokens": 43581.0,
      "step": 25
    },
    {
      "entropy": 1.912336548268795,
      "epoch": 0.08888888888888889,
      "grad_norm": 41.25,
      "learning_rate": 5.7988165680473375e-06,
      "loss": 2.3563003540039062,
      "mean_token_accuracy": 0.5931275233626365,
      "num_tokens": 86889.0,
      "step": 50
    },
    {
      "entropy": 2.1532553315162657,
      "epoch": 0.13333333333333333,
      "grad_norm": 31.125,
      "learning_rate": 8.757396449704143e-06,
      "loss": 2.159627685546875,
      "mean_token_accuracy": 0.5959913417696953,
      "num_tokens": 130613.0,
      "step": 75
    },
    {
      "entropy": 1.9658770501613616,
      "epoch": 0.17777777777777778,
      "grad_norm": 30.0,
      "learning_rate": 1.1715976331360948e-05,
      "loss": 1.9642219543457031,
      "mean_token_accuracy": 0.6274332204461097,
      "num_tokens": 174274.0,
      "step": 100
    },
    {
      "entropy": 1.9181486976146698,
      "epoch": 0.2222222222222222,
      "grad_norm": 23.875,
      "learning_rate": 1.4674556213017752e-05,
      "loss": 1.9006834411621094,
      "mean_token_accuracy": 0.6362838646769524,
      "num_tokens": 217619.0,
      "step": 125
    },
    {
      "entropy": 1.8227408236265183,
      "epoch": 0.26666666666666666,
      "grad_norm": 29.5,
      "learning_rate": 1.7633136094674557e-05,
      "loss": 1.834572296142578,
      "mean_token_accuracy": 0.645097343325615,
      "num_tokens": 261407.0,
      "step": 150
    },
    {
      "entropy": 1.8350859326124191,
      "epoch": 0.3111111111111111,
      "grad_norm": 25.625,
      "learning_rate": 1.999946602771351e-05,
      "loss": 1.8147303771972656,
      "mean_token_accuracy": 0.6461810152232647,
      "num_tokens": 305402.0,
      "step": 175
    },
    {
      "entropy": 1.753722729086876,
      "epoch": 0.35555555555555557,
      "grad_norm": 31.5,
      "learning_rate": 1.9980782984658682e-05,
      "loss": 1.758114013671875,
      "mean_token_accuracy": 0.653534509241581,
      "num_tokens": 348998.0,
      "step": 200
    },
    {
      "entropy": 1.7685393619537353,
      "epoch": 0.4,
      "grad_norm": 23.75,
      "learning_rate": 1.9935458325191365e-05,
      "loss": 1.7483135986328124,
      "mean_token_accuracy": 0.6603708426654339,
      "num_tokens": 392837.0,
      "step": 225
    },
    {
      "entropy": 1.7378248673677446,
      "epoch": 0.4444444444444444,
      "grad_norm": 26.5,
      "learning_rate": 1.9863613034027224e-05,
      "loss": 1.7749085998535157,
      "mean_token_accuracy": 0.65810049533844,
      "num_tokens": 436281.0,
      "step": 250
    },
    {
      "entropy": 1.6766968676447869,
      "epoch": 0.4888888888888889,
      "grad_norm": 23.25,
      "learning_rate": 1.9765438887171327e-05,
      "loss": 1.668856658935547,
      "mean_token_accuracy": 0.6644712510704994,
      "num_tokens": 480451.0,
      "step": 275
    },
    {
      "entropy": 1.669287955760956,
      "epoch": 0.5333333333333333,
      "grad_norm": 20.625,
      "learning_rate": 1.9641197940012136e-05,
      "loss": 1.648075714111328,
      "mean_token_accuracy": 0.6689342276751995,
      "num_tokens": 524418.0,
      "step": 300
    },
    {
      "entropy": 1.587462457716465,
      "epoch": 0.5777777777777777,
      "grad_norm": 21.625,
      "learning_rate": 1.9491221827819348e-05,
      "loss": 1.5674880981445312,
      "mean_token_accuracy": 0.6790469540655613,
      "num_tokens": 568093.0,
      "step": 325
    },
    {
      "entropy": 1.5877643951773643,
      "epoch": 0.6222222222222222,
      "grad_norm": 23.5,
      "learning_rate": 1.9315910880512792e-05,
      "loss": 1.5833602905273438,
      "mean_token_accuracy": 0.67303916528821,
      "num_tokens": 611875.0,
      "step": 350
    },
    {
      "entropy": 1.5628123182058333,
      "epoch": 0.6666666666666666,
      "grad_norm": 19.875,
      "learning_rate": 1.911573305406528e-05,
      "loss": 1.5389964294433593,
      "mean_token_accuracy": 0.6916262510418892,
      "num_tokens": 655258.0,
      "step": 375
    },
    {
      "entropy": 1.5605447962880135,
      "epoch": 0.7111111111111111,
      "grad_norm": 25.25,
      "learning_rate": 1.8891222681391853e-05,
      "loss": 1.5674354553222656,
      "mean_token_accuracy": 0.6811535279452801,
      "num_tokens": 698915.0,
      "step": 400
    },
    {
      "entropy": 1.5302811539173127,
      "epoch": 0.7555555555555555,
      "grad_norm": 21.0,
      "learning_rate": 1.8642979046059595e-05,
      "loss": 1.5149029541015624,
      "mean_token_accuracy": 0.6928434443473815,
      "num_tokens": 742608.0,
      "step": 425
    },
    {
      "entropy": 1.4984435787796975,
      "epoch": 0.8,
      "grad_norm": 20.5,
      "learning_rate": 1.8371664782625287e-05,
      "loss": 1.4783544921875,
      "mean_token_accuracy": 0.6922086542844772,
      "num_tokens": 785924.0,
      "step": 450
    },
    {
      "entropy": 1.4762829530239105,
      "epoch": 0.8444444444444444,
      "grad_norm": 22.875,
      "learning_rate": 1.8078004107870797e-05,
      "loss": 1.4516114807128906,
      "mean_token_accuracy": 0.702991289794445,
      "num_tokens": 829069.0,
      "step": 475
    },
    {
      "entropy": 1.5141191801428795,
      "epoch": 0.8888888888888888,
      "grad_norm": 20.75,
      "learning_rate": 1.7762780887657576e-05,
      "loss": 1.5113209533691405,
      "mean_token_accuracy": 0.6913984973728656,
      "num_tokens": 872824.0,
      "step": 500
    },
    {
      "entropy": 1.4147853454947472,
      "epoch": 0.9333333333333333,
      "grad_norm": 25.75,
      "learning_rate": 1.74268365445604e-05,
      "loss": 1.394475555419922,
      "mean_token_accuracy": 0.712423055768013,
      "num_tokens": 916212.0,
      "step": 525
    },
    {
      "entropy": 1.4362201416492462,
      "epoch": 0.9777777777777777,
      "grad_norm": 21.25,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 1.4004106140136718,
      "mean_token_accuracy": 0.7106147593259812,
      "num_tokens": 959695.0,
      "step": 550
    },
    {
      "epoch": 1.0,
      "eval_entropy": 1.3886345291137696,
      "eval_loss": 1.3774118423461914,
      "eval_mean_token_accuracy": 0.7105021185874939,
      "eval_num_tokens": 981504.0,
      "eval_runtime": 28.7124,
      "eval_samples_per_second": 34.828,
      "eval_steps_per_second": 4.354,
      "step": 563
    },
    {
      "entropy": 1.3702899970266285,
      "epoch": 1.0213333333333334,
      "grad_norm": 24.375,
      "learning_rate": 1.6696424339928153e-05,
      "loss": 1.2998457336425782,
      "mean_token_accuracy": 0.7251553581077226,
      "num_tokens": 1002442.0,
      "step": 575
    },
    {
      "entropy": 1.2198174016177654,
      "epoch": 1.0657777777777777,
      "grad_norm": 19.0,
      "learning_rate": 1.6303906161279554e-05,
      "loss": 1.1578231048583985,
      "mean_token_accuracy": 0.7492367821931839,
      "num_tokens": 1045974.0,
      "step": 600
    },
    {
      "entropy": 1.2214804862439632,
      "epoch": 1.1102222222222222,
      "grad_norm": 21.5,
      "learning_rate": 1.5894561021248535e-05,
      "loss": 1.1222823333740235,
      "mean_token_accuracy": 0.7596075300872326,
      "num_tokens": 1089691.0,
      "step": 625
    },
    {
      "entropy": 1.230571178793907,
      "epoch": 1.1546666666666667,
      "grad_norm": 18.25,
      "learning_rate": 1.5469481581224274e-05,
      "loss": 1.1252061462402343,
      "mean_token_accuracy": 0.7554993903636933,
      "num_tokens": 1133544.0,
      "step": 650
    },
    {
      "entropy": 1.2035815131664276,
      "epoch": 1.199111111111111,
      "grad_norm": 17.75,
      "learning_rate": 1.5029802502024788e-05,
      "loss": 1.1525257110595704,
      "mean_token_accuracy": 0.7539819921553135,
      "num_tokens": 1176944.0,
      "step": 675
    },
    {
      "entropy": 1.1947351975739002,
      "epoch": 1.2435555555555555,
      "grad_norm": 19.5,
      "learning_rate": 1.4576697415156818e-05,
      "loss": 1.1312750244140626,
      "mean_token_accuracy": 0.755437722504139,
      "num_tokens": 1220536.0,
      "step": 700
    },
    {
      "entropy": 1.2491805362701416,
      "epoch": 1.288,
      "grad_norm": 18.625,
      "learning_rate": 1.4111375790051511e-05,
      "loss": 1.1570246124267578,
      "mean_token_accuracy": 0.7577420398592949,
      "num_tokens": 1264076.0,
      "step": 725
    },
    {
      "entropy": 1.2654732164740563,
      "epoch": 1.3324444444444445,
      "grad_norm": 20.875,
      "learning_rate": 1.3635079705638298e-05,
      "loss": 1.1855599212646484,
      "mean_token_accuracy": 0.751485439389944,
      "num_tokens": 1308034.0,
      "step": 750
    },
    {
      "entropy": 1.2152655026316643,
      "epoch": 1.3768888888888888,
      "grad_norm": 20.25,
      "learning_rate": 1.3149080534874519e-05,
      "loss": 1.1257845306396483,
      "mean_token_accuracy": 0.7632686151564121,
      "num_tokens": 1351646.0,
      "step": 775
    },
    {
      "entropy": 1.193697052896023,
      "epoch": 1.4213333333333333,
      "grad_norm": 24.75,
      "learning_rate": 1.2654675551080724e-05,
      "loss": 1.121256103515625,
      "mean_token_accuracy": 0.7563131806254387,
      "num_tokens": 1395047.0,
      "step": 800
    },
    {
      "entropy": 1.1570866696536541,
      "epoch": 1.4657777777777778,
      "grad_norm": 22.25,
      "learning_rate": 1.2153184465140413e-05,
      "loss": 1.0696445465087892,
      "mean_token_accuracy": 0.765836420059204,
      "num_tokens": 1438280.0,
      "step": 825
    },
    {
      "entropy": 1.151665413379669,
      "epoch": 1.5102222222222221,
      "grad_norm": 17.75,
      "learning_rate": 1.164594590280734e-05,
      "loss": 1.0663511657714844,
      "mean_token_accuracy": 0.7644271370768547,
      "num_tokens": 1481978.0,
      "step": 850
    },
    {
      "entropy": 1.2120365902781487,
      "epoch": 1.5546666666666666,
      "grad_norm": 17.5,
      "learning_rate": 1.1134313831523547e-05,
      "loss": 1.1270946502685546,
      "mean_token_accuracy": 0.7585732811689376,
      "num_tokens": 1525597.0,
      "step": 875
    },
    {
      "entropy": 1.151113938242197,
      "epoch": 1.5991111111111111,
      "grad_norm": 21.5,
      "learning_rate": 1.0619653946285948e-05,
      "loss": 1.0662925720214844,
      "mean_token_accuracy": 0.7686286368966102,
      "num_tokens": 1568743.0,
      "step": 900
    },
    {
      "entropy": 1.162338248193264,
      "epoch": 1.6435555555555554,
      "grad_norm": 20.625,
      "learning_rate": 1.0103340024208674e-05,
      "loss": 1.0983226013183593,
      "mean_token_accuracy": 0.7634404581785202,
      "num_tokens": 1612238.0,
      "step": 925
    },
    {
      "entropy": 1.1956017450988292,
      "epoch": 1.688,
      "grad_norm": 18.625,
      "learning_rate": 9.586750257511868e-06,
      "loss": 1.1054475402832031,
      "mean_token_accuracy": 0.7624230819940567,
      "num_tokens": 1656115.0,
      "step": 950
    },
    {
      "entropy": 1.1524758899211884,
      "epoch": 1.7324444444444445,
      "grad_norm": 20.875,
      "learning_rate": 9.0712635747253e-06,
      "loss": 1.063654327392578,
      "mean_token_accuracy": 0.7672438740730285,
      "num_tokens": 1700143.0,
      "step": 975
    },
    {
      "entropy": 1.1726488883793353,
      "epoch": 1.7768888888888887,
      "grad_norm": 18.25,
      "learning_rate": 8.558255959926533e-06,
      "loss": 1.0800904846191406,
      "mean_token_accuracy": 0.767799711227417,
      "num_tokens": 1743486.0,
      "step": 1000
    },
    {
      "entropy": 1.1595972776412964,
      "epoch": 1.8213333333333335,
      "grad_norm": 22.125,
      "learning_rate": 8.04909677983872e-06,
      "loss": 1.100532989501953,
      "mean_token_accuracy": 0.7661877936124801,
      "num_tokens": 1787577.0,
      "step": 1025
    },
    {
      "entropy": 1.1869910764694214,
      "epoch": 1.8657777777777778,
      "grad_norm": 18.625,
      "learning_rate": 7.545145128592009e-06,
      "loss": 1.1068744659423828,
      "mean_token_accuracy": 0.7658174222707749,
      "num_tokens": 1831146.0,
      "step": 1050
    },
    {
      "entropy": 1.1678849336504937,
      "epoch": 1.9102222222222223,
      "grad_norm": 20.25,
      "learning_rate": 7.0477461999055365e-06,
      "loss": 1.104344940185547,
      "mean_token_accuracy": 0.7647965340316296,
      "num_tokens": 1875315.0,
      "step": 1075
    },
    {
      "entropy": 1.2021603246033192,
      "epoch": 1.9546666666666668,
      "grad_norm": 20.75,
      "learning_rate": 6.558227696373617e-06,
      "loss": 1.1086157989501952,
      "mean_token_accuracy": 0.7716675865650177,
      "num_tokens": 1918868.0,
      "step": 1100
    },
    {
      "entropy": 1.1465536914765835,
      "epoch": 1.999111111111111,
      "grad_norm": 24.0,
      "learning_rate": 6.077896285440874e-06,
      "loss": 1.0573983001708984,
      "mean_token_accuracy": 0.7682515250146389,
      "num_tokens": 1962187.0,
      "step": 1125
    },
    {
      "epoch": 2.0,
      "eval_entropy": 1.1745851793289184,
      "eval_loss": 1.2738397121429443,
      "eval_mean_token_accuracy": 0.730101492881775,
      "eval_num_tokens": 1963008.0,
      "eval_runtime": 28.609,
      "eval_samples_per_second": 34.954,
      "eval_steps_per_second": 4.369,
      "step": 1126
    },
    {
      "entropy": 1.101325544167538,
      "epoch": 2.042666666666667,
      "grad_norm": 15.375,
      "learning_rate": 5.608034111526298e-06,
      "loss": 0.9211092376708985,
      "mean_token_accuracy": 0.8091437579417715,
      "num_tokens": 2004400.0,
      "step": 1150
    },
    {
      "entropy": 1.0406066159904004,
      "epoch": 2.087111111111111,
      "grad_norm": 20.5,
      "learning_rate": 5.149895373606405e-06,
      "loss": 0.9091825103759765,
      "mean_token_accuracy": 0.8042202299833298,
      "num_tokens": 2048143.0,
      "step": 1175
    },
    {
      "entropy": 1.0808034761250018,
      "epoch": 2.1315555555555554,
      "grad_norm": 18.0,
      "learning_rate": 4.704702977392914e-06,
      "loss": 0.9504242706298828,
      "mean_token_accuracy": 0.7970917990803719,
      "num_tokens": 2091837.0,
      "step": 1200
    },
    {
      "entropy": 1.0659918001294135,
      "epoch": 2.176,
      "grad_norm": 20.75,
      "learning_rate": 4.2736452710412645e-06,
      "loss": 0.9259893798828125,
      "mean_token_accuracy": 0.8023737239837646,
      "num_tokens": 2135628.0,
      "step": 1225
    },
    {
      "entropy": 1.0154806707799435,
      "epoch": 2.2204444444444444,
      "grad_norm": 20.0,
      "learning_rate": 3.857872873103322e-06,
      "loss": 0.8762107086181641,
      "mean_token_accuracy": 0.8123826086521149,
      "num_tokens": 2178782.0,
      "step": 1250
    },
    {
      "entropy": 1.0083947947621346,
      "epoch": 2.2648888888888887,
      "grad_norm": 21.375,
      "learning_rate": 3.4584956011913693e-06,
      "loss": 0.8710794830322266,
      "mean_token_accuracy": 0.8071741372346878,
      "num_tokens": 2222303.0,
      "step": 1275
    },
    {
      "entropy": 1.0441110938787461,
      "epoch": 2.3093333333333335,
      "grad_norm": 18.75,
      "learning_rate": 3.0765795095517026e-06,
      "loss": 0.9363404846191407,
      "mean_token_accuracy": 0.7974440288543702,
      "num_tokens": 2266083.0,
      "step": 1300
    },
    {
      "entropy": 1.0385051570832728,
      "epoch": 2.3537777777777777,
      "grad_norm": 20.625,
      "learning_rate": 2.713144043455388e-06,
      "loss": 0.9435790252685546,
      "mean_token_accuracy": 0.8013337072730065,
      "num_tokens": 2309679.0,
      "step": 1325
    },
    {
      "entropy": 1.041314078271389,
      "epoch": 2.398222222222222,
      "grad_norm": 18.25,
      "learning_rate": 2.369159318001937e-06,
      "loss": 0.9011815643310547,
      "mean_token_accuracy": 0.8010741752386094,
      "num_tokens": 2353394.0,
      "step": 1350
    },
    {
      "entropy": 1.045836333334446,
      "epoch": 2.4426666666666668,
      "grad_norm": 20.625,
      "learning_rate": 2.045543528599607e-06,
      "loss": 0.9371556091308594,
      "mean_token_accuracy": 0.8022435599565506,
      "num_tokens": 2397064.0,
      "step": 1375
    },
    {
      "entropy": 1.02559718221426,
      "epoch": 2.487111111111111,
      "grad_norm": 16.875,
      "learning_rate": 1.743160500034443e-06,
      "loss": 0.9111165618896484,
      "mean_token_accuracy": 0.8056430542469024,
      "num_tokens": 2440899.0,
      "step": 1400
    },
    {
      "entropy": 0.9850550366938114,
      "epoch": 2.5315555555555553,
      "grad_norm": 18.25,
      "learning_rate": 1.4628173806703594e-06,
      "loss": 0.8460595703125,
      "mean_token_accuracy": 0.814309296309948,
      "num_tokens": 2484527.0,
      "step": 1425
    },
    {
      "entropy": 1.0671714194118977,
      "epoch": 2.576,
      "grad_norm": 16.5,
      "learning_rate": 1.2052624879351105e-06,
      "loss": 0.9366078186035156,
      "mean_token_accuracy": 0.7982332086563111,
      "num_tokens": 2528184.0,
      "step": 1450
    },
    {
      "entropy": 1.0057188278436662,
      "epoch": 2.6204444444444444,
      "grad_norm": 18.25,
      "learning_rate": 9.711833108431234e-07,
      "loss": 0.8691239929199219,
      "mean_token_accuracy": 0.8100746890902519,
      "num_tokens": 2571871.0,
      "step": 1475
    },
    {
      "entropy": 1.0419890847057105,
      "epoch": 2.664888888888889,
      "grad_norm": 18.25,
      "learning_rate": 7.612046748871327e-07,
      "loss": 0.8980624389648437,
      "mean_token_accuracy": 0.8050528278946877,
      "num_tokens": 2615338.0,
      "step": 1500
    },
    {
      "entropy": 1.0672108069062234,
      "epoch": 2.7093333333333334,
      "grad_norm": 21.25,
      "learning_rate": 5.758870741969635e-07,
      "loss": 0.9245491790771484,
      "mean_token_accuracy": 0.8008973941206932,
      "num_tokens": 2659309.0,
      "step": 1525
    },
    {
      "entropy": 1.0332248322665691,
      "epoch": 2.7537777777777777,
      "grad_norm": 17.25,
      "learning_rate": 4.1572517541747294e-07,
      "loss": 0.9085790252685547,
      "mean_token_accuracy": 0.8022564603388309,
      "num_tokens": 2703096.0,
      "step": 1550
    },
    {
      "entropy": 1.0293236848711969,
      "epoch": 2.7982222222222224,
      "grad_norm": 18.125,
      "learning_rate": 2.811464972992195e-07,
      "loss": 0.8811993408203125,
      "mean_token_accuracy": 0.8039135798811913,
      "num_tokens": 2746647.0,
      "step": 1575
    },
    {
      "entropy": 1.0229834078252316,
      "epoch": 2.8426666666666667,
      "grad_norm": 25.625,
      "learning_rate": 1.7251026952640583e-07,
      "loss": 0.90928466796875,
      "mean_token_accuracy": 0.806032055914402,
      "num_tokens": 2790363.0,
      "step": 1600
    },
    {
      "entropy": 1.051690325587988,
      "epoch": 2.887111111111111,
      "grad_norm": 21.75,
      "learning_rate": 9.010647382825421e-08,
      "loss": 0.9022364807128906,
      "mean_token_accuracy": 0.8011369660496712,
      "num_tokens": 2834043.0,
      "step": 1625
    },
    {
      "entropy": 0.9921600778400897,
      "epoch": 2.9315555555555557,
      "grad_norm": 21.75,
      "learning_rate": 3.4155069933301535e-08,
      "loss": 0.8524517822265625,
      "mean_token_accuracy": 0.8153755596280098,
      "num_tokens": 2877282.0,
      "step": 1650
    },
    {
      "entropy": 1.0421569736301899,
      "epoch": 2.976,
      "grad_norm": 20.625,
      "learning_rate": 4.80540843283972e-09,
      "loss": 0.93547607421875,
      "mean_token_accuracy": 0.8020336532592773,
      "num_tokens": 2920814.0,
      "step": 1675
    },
    {
      "epoch": 3.0,
      "eval_entropy": 1.1009210176467896,
      "eval_loss": 1.3000445365905762,
      "eval_mean_token_accuracy": 0.7269604291915893,
      "eval_num_tokens": 2944512.0,
      "eval_runtime": 28.9685,
      "eval_samples_per_second": 34.52,
      "eval_steps_per_second": 4.315,
      "step": 1689
    }
  ],
  "logging_steps": 25,
  "max_steps": 1689,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7572413980065792.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}